Featured Engineering
A curated selection of high-performance systems and full-stack applications. Exploring the intersection of AI, scalable infrastructure, and intuitive user experiences.
PrepMeUp
A student exam preparation and evaluation system with notes, mock tests, and exam evaluation.
The Challenge
Traditional interview prep lacks dynamic, real-time feedback mimicking actual human interaction, leaving candidates unprepared for the unpredictability of live interviews.
The Approach
Architected a low-latency WebRTC streaming pipeline integrated with custom LLMs to analyze speech, tone, and technical accuracy in real-time. Deployed a microservices architecture for scalable concurrent mock sessions.
Key Results
- trending_up 40% increase in user reported placement rates.
- speed Sub-200ms latency for AI feedback generation.
export class WorkflowEngine {
private queue: RedisQueue;
private workers: WorkerPool;
constructor(config: EngineConfig) {
this.queue = new RedisQueue(config.redis);
this.workers = new WorkerPool(config.maxWorkers);
}
async processTask(taskId: string): Promise<Result> {
const task = await this.queue.dequeue(taskId);
if (!task) {
throw new Error('Task not found in queue');
}
try {
this.metrics.recordStart(taskId);
const worker = await this.workers.acquire();
const result = await worker.execute(task);
this.metrics.recordSuccess(taskId);
return result;
} catch (error) {
this.handleFailure(task, error);
}
}
}
FlowQ
A hydration tracker and reminder system.
The Challenge
Existing task runners struggled with complex DAG (Directed Acyclic Graph) dependencies at high concurrency, leading to race conditions and bottlenecked execution pipelines.
The Approach
Implemented a custom distributed queue system using Redis and a worker pool architecture in Go. Designed an event-driven reconciliation loop to guarantee at-least-once execution and topological sorting for DAG resolution.
Key Results
- memory Processed 1M+ tasks/day with 99.99% uptime.
- compress Reduced memory footprint by 60% compared to legacy solutions.
Kavach
A Digital Identity Protection and Security System for Women.
The Challenge
Providing military-grade security for civilian communication while maintaining a seamless, zero-latency user experience on mobile devices with limited computational power.
The Approach
Developed the core cryptographic engine in Rust for memory safety and execution speed, compiled to WebAssembly for cross-platform deployment. Implemented a zero-knowledge proof authentication flow.
Key Results
- shield_locked Achieved AES-256-GCM encryption with <5ms overhead.
- policy Successfully audited by third-party security firm with zero critical vulnerabilities.
[INFO] Negotiating TLS 1.3 handshake... OK
[INFO] Generating ephemeral keypairs... OK
[SUCCESS] End-to-end encryption established.
[INFO] Awaiting data stream...
System Architecture & Database Design
Contact-Book-Dashboard
Flask • SQL • RedisA high-throughput contact management dashboard engineered to handle massive datasets with minimal query latency.
- > Optimized SQL indexes for 50,000+ record handling.
- > Achieved sub-100ms API response times.
- > Implemented Redis caching layer for frequent read queries.
Inventory Management
React • GraphQL • PostgreSQLA comprehensive ERP module for tracking physical assets, managing supply chains, and automating reorder workflows.
- > Reduced operational costs by 35% via automated workflows.
- > Migrated REST endpoints to GraphQL to resolve over-fetching.
- > Built real-time inventory sync using WebSockets.